This is a quickly written search engine put together on Good Friday (10/4/98) between chocolate Easter eggs! It will probably be revised over the next few weeks.

Current Features...

• The search strings are compared against a listing of files in the /pub/ hierarchy. Matches are on substrings and are case-insensitive.
• You can use multiple search keys. All search strings are AND'ed together. eg. to search for a Macintosh version of Netscape, you could search for "mac netscape".
• You can enter as many space-separated search strings as you like.
• Results are returned in directory order, which makes more sense then pure alphabetic.
• FYI the search dataset is the /pub/ls-lR file and the program is written in Perl5 using mod_perl with Apache.

Current Bugs...

• Sometimes the script fails with "Internal Server Error". The real error is "Can't undef active subroutine". I'm not sure why yet! I'm sure I'm missing something in mod_perl.

Things Todo...

• OR'ing of search keys. Initially just simplistic OR'ing of two or more search strings, eg. "unix ftp mail" to return all filenames which have any of those strings.
• Sorting of results based on date, size and name ala Apache 1.3.
• Search stats - search time, number of matches, total file size, etc.
• Allow choice about using case-insensitive and sub-string matching.
• Directory statistics: total size of directory, last modified date.
• Limit the number of entries returned. Probably allow a user-defined "maximum number of hits". Defaulting to something like 100.
• Top 50 new files. Files which are new in the past N days or weeks.
• Look at regular expression matching (this sort of works now, but I've limited the search strings so that the characters used for regex are invalid, until I can determine the how safe it is!).