home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
IRIX Installation Tools & Overlays 2002 November
/
SGI IRIX Installation Tools & Overlays 2002 November - Disc 4.iso
/
dist
/
sgsearch.idb
/
usr
/
sgsearch
/
conf
/
sgbot.conf.z
/
sgbot.conf
Wrap
Text File
|
2002-10-15
|
2KB
|
110 lines
## SGBot configuration file
## version 0.8
##
## Alan Braverman <alanb@sgi.com>
## December 6, 1996
## Index Directory
##
## This is the directory into which SGBot will write the index files,
## so this directory must be writable by the user running SGBot, of
## course.
indexDir /usr/sgsearch/index/
## Temporary Directory
##
## This is a directory into which SGBot will stick (and then delete)
## temporary files.
tmpDir /tmp/
## Index Binary Directory
##
## This is the directory in which SGBot can find the other executables
## it needs for indexing (sgreader, sgdct, sginverter, sgmerge)
indexBinDir /usr/sbin/
## Include Extension
##
## The includeExtension variables tell SGBot what file extensions it
## should assume are HTML or ASCII text.
includeExtension asci
includeExtension ascii
includeExtension faq
includeExtension html
includeExtension htm
includeExtension info
includeExtension readme
includeExtension text
includeExtension txt
## Exclude Scheme
##
## The excludeScheme variables tells SGBot what URL schemes (or
## protocols) to ignore.
excludeScheme gopher
excludeScheme https
excludeScheme mail
excludeScheme mailto
excludeScheme news
excludeScheme nntp
excludeScheme rlogin
excludeScheme shttp
excludeScheme telnet
excludeScheme wais
## Allow Host
##
## The allowHost variables tell SGBot what hostnames are allowed to be
## indexed. All others will be ignored. Wildcards may be used to specify
## subdomains. SGBot is not yet very smart when it comes to partial
## hostnames.
allowHost void
# allowHost void.esd
# allowHost void.esd.sgi.com
# allowHost *.esd.sgi.com
## URL
##
## The url variables tell SGBot at what URLs to begin its search.
## Hostnames found in these URLs are automatically added to the allowHost
## list.
url http://void
# url http://foo
# url http://bar
## Exclude URLs
##
## The excludeUrl variables tell SGBot what URLs to ignore. The
## wildcard character (*) may be appended to the end of any URL.
excludeUrl http://void.esd.sgi.com/not-me.html
excludeUrl http://void.esd.sgi.com/private/*
## Filename Base
##
## The filenameBase variable tells SGBot what to name all the files
## it dumps out, like sgbot.dct, sgbot.inv, etc.
filenameBase sgbot
## Verbose Output
##
## The verbose flag tells SGBot to dump extra log information.
## Uncomment this line for verbose output
# verbose true
## Time Out
##
## The timeOut variable sets the time (in seconds) that SGBot will
## wait for a slow document
timeOut 30