#!/usr/bin/tclsh
-# $Id: robot.tcl,v 1.23 2001/10/31 08:51:49 adam Exp $
+# $Id: robot.tcl,v 1.24 2001/11/07 11:30:52 adam Exp $
#
proc RobotFileNext1 {area lead} {
# puts "RobotFileNext1 area=$area lead=$lead"
RobotWriteMetadata $url $out
RobotFileClose $out
- if {[file isdirectory flat]} {
- regsub -all {/} $URL($url,hostport).$URL($url,path) {.} fname
- set out [open "flat/$fname" w]
- RobotWriteMetadata $url $out
- close $out
- }
RobotFileUnlink unvisited $URL($url,hostport) $URL($url,path)
}
# consider type
if {[lindex $l 1] != $type} continue
# consider mask (! negates)
- set mask [lindex $l 2]
- if {[string index $mask 0] == "!"} {
- set mask [string range $mask 1 end]
- if {[string match $mask $this]} continue
- } else {
- if {![string match $mask $this]} continue
+ set masks [lindex $l 2]
+ set ok 0
+ foreach mask $masks {
+ if {$debuglevel > 4} {
+ puts "consider single mask $mask"
+ }
+ if {[string index $mask 0] == "!"} {
+ set mask [string range $mask 1 end]
+ if {[string match $mask $this]} continue
+ } else {
+ if {![string match $mask $this]} continue
+ }
+ set ok 1
+ }
+ if {$debuglevel > 4} {
+ puts "ok = $ok"
}
+ if {!$ok} continue
# OK, we have a match
if {[lindex $l 0] == "allow"} {
if {$debuglevel > 3} {
- puts "CHECKRULE MATH OK"
+ puts "CHECKRULE MATCH OK"
}
return 1
} else {
}
}
if {$debuglevel > 3} {
- puts "CHECKRULE MATH OK"
+ puts "CHECKRULE MATCH OK"
}
return 1
}
-# sample rules $Id: rules,v 1.2 2001/10/30 08:29:54 adam Exp $
+# sample rules $Id: rules,v 1.3 2001/11/07 11:30:52 adam Exp $
url http://www.indexdata.dk
-allow url http://www.indexdata.dk/software*
-allow url http://www.indexdata.dk/zap*
+allow url {http://www.indexdata.dk/software* http://www.indexdata.dk/zap*}
allow url http://www.indexdata.dk/yaz*
allow url http://www.indexdata.dk/
deny url http://www.indexdata.dk/*
allow mime text/html
allow mime application/pdf
-deny mime text/plain
+allow mime text/plain
deny mime *
-set maxdistance 4
+set maxdistance 3
+
+set debuglevel 1