home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Collection of Internet
/
Collection of Internet.iso
/
infosrvr
/
dev
/
www_talk.930
/
000097_timbl _Mon May 25 16:40:01 1992.msg
< prev
next >
Wrap
Internet Message Format
|
1994-01-24
|
1KB
Return-Path: <timbl>
Received: by nxoc01.cern.ch (NeXT-1.0 (From Sendmail 5.52)/NeXT-2.0)
id AA18996; Mon, 25 May 92 16:40:01 GMT+0200
Date: Mon, 25 May 92 16:40:01 GMT+0200
From: timbl (Tim Berners-Lee)
Message-Id: <9205251440.AA18996@ nxoc01.cern.ch >
Received: by NeXT Mailer (1.62)
To: pflynn@curia.ucc.ie (Peter Flynn)
Subject: Re: search engines & views
Cc: www-talk@nxoc01.cern.ch
> All you do is map the parameters of the virtual search engine
> onto a document name -- like
>
> /INDEX/full-text/tryhard/depth=5/boolean
>
> Does this mean (at a primitive level) you could code a grep command
> as a document name?
>
Yes -- sure. Its a question of writing down the
algorithm. In perl, I'm sure its a cinch ... you could
also do it with sh and sed :-( but basically for example
you need to take say
/grep/mydir/i?joe+bloggs
and turn that into
grep -l -i "(joe)|(bloggs)" | awk -f ls2html.awk
where ls2html.awk looks something like:
BEGIN { print "Select one of:\n<MENU>" }
{ printf "<LI><A HREF=./%s> %s</A>\n", $1, $1 }
END { print "</MENU>" }
The awk generates the HTML for a menu. I guess you could use awk in fact to
generate the grep command too. But these are just ideas. or are you using VMS?
Yes, you could probably do it with DCL and SEARCH.
Of course if you can handle C, then hack the sample httpd.
> ///Peter
Tim