#!/usr/bin/tclsh
-# $Id: robot.tcl,v 1.9 2000/12/11 17:11:03 adam Exp $
+# $Id: robot.tcl,v 1.18 2001/06/07 08:17:00 adam Exp $
#
proc RobotFileNext1 {area lead} {
- puts "RobotFileNext1 area=$area lead=$lead"
+ # puts "RobotFileNext1 area=$area lead=$lead"
if {[catch {set ns [glob ${area}/*]}]} {
return {}
}
return {}
}
-proc RobotFileWait {} {
- global robotSeq
- set robotSeq 0
+proc RobotWriteRecord {outf fromurl distance} {
+ puts $outf "<zmbot>"
+ puts $outf "<distance>"
+ puts $outf $distance
+ puts $outf "</distance>"
+ puts $outf "<fromurl>"
+ puts $outf $fromurl
+ puts $outf "</fromurl>"
+ puts $outf "</zmbot>"
+}
+
+proc RobotReadRecord {inf fromurlx distancex} {
+ upvar $fromurlx fromurl
+ upvar $distancex distance
+ gets $inf
+ gets $inf
+ set distance [string trim [gets $inf]]
+ # puts "got distance = $distance"
+ gets $inf
+ gets $inf
+ set fromurl [string trim [gets $inf]]
}
proc RobotFileNext {area} {
- global robotSeq
- puts "RobotFileNext robotSeq=$robotSeq"
- if {[catch {set ns [glob ${area}/*]}]} {
- return {}
+ global robotSeq global idleTime ns
+
+ # puts "RobotFileNext robotSeq=$robotSeq"
+ if {$robotSeq < 0} {
+ return {}
+ }
+ if {$robotSeq == 0} {
+ if {[catch {set ns [glob ${area}/*]}]} {
+ return {}
+ }
}
set off [string length $area]
incr off
-
set n [lindex $ns $robotSeq]
if {![string length $n]} {
+ set robotSeq -1
flush stdout
puts "------------ N E X T R O U N D --------"
- set robotSeq -1
- after 60000 RobotFileWait
- vwait robotSeq
-
- set n [lindex $ns $robotSeq]
- if {![string length $n]} {
- puts "robotSeq = $robotSeq"
- puts "ns=$ns"
- puts "no more work at index"
- return {}
- }
+ return wait
}
incr robotSeq
if {[file isfile $n/frobots.txt]} {
proc RobotFileExist {area host path} {
- puts "RobotFileExist begin"
- puts "area=$area host=$host path=$path"
+ # puts "RobotFileExist begin area=$area host=$host path=$path"
set lpath [split $path /]
set l [llength $lpath]
incr l -1
set t [lindex $lpath $l]
incr l -1
set npath $area/$host[join [lrange $lpath 0 $l] /d]/f$t
- puts "npath=$npath"
- puts "RobotFileExist end"
+ # puts "RobotFileExist end npath=$npath"
return [file exists $npath]
}
proc RobotFileUnlink {area host path} {
- puts "RobotFileUnlink begin"
- puts "area=$area host=$host path=$path"
+ # puts "RobotFileUnlink begin"
+ # puts "area=$area host=$host path=$path"
set lpath [split $path /]
set l [llength $lpath]
incr l -1
set t [lindex $lpath $l]
incr l -1
set npath $area/$host[join [lrange $lpath 0 $l] /d]/f$t
- puts "npath=$npath"
+ # puts "npath=$npath"
set comp [split $npath /]
set l [llength $comp]
incr l -1
if {![catch {glob $path/*}]} return
exec rmdir ./$path
}
- puts "RobotFileUnlink end"
+ # puts "RobotFileUnlink end"
}
proc RobotFileClose {out} {
if {![info exists workdir]} {
return stdout
}
- puts "RobotFileOpen orgPwd=$orgPwd area=$area host=$host path=$path mode=$mode"
+ #puts "RobotFileOpen orgPwd=$orgPwd area=$area host=$host path=$path mode=$mode"
if {[string compare $orgPwd $workdir]} {
puts "ooops. RobotFileOpen failed"
puts "workdir = $workdir"
}
if {[catch {cd ./$d}]} {
exec mkdir $d
- puts "creating $d"
cd ./$d
if {![string compare $area unvisited] && $i == 1 && $mode == "w"} {
set out [open frobots.txt w]
if {[string length $d]} {
if {[file isdirectory $d]} {
set out [open $d/f $mode]
- puts "1"
} else {
set out [open f$d $mode]
- puts "2"
}
} else {
set out [open f $mode]
- puts "3"
}
cd $orgPwd
- #puts "RobotFileStop"
return $out
}
-proc RobotRestart {sock} {
- global URL
- global robotMoreWork
-
+proc RobotRR {} {
+ global robotSeq robotsRunning
+
+ incr robotsRunning -1
+ while {$robotsRunning} {
+ vwait robotsRunning
+ }
+ set robotSeq 0
+ RobotStart
+}
+
+proc RobotRestart {url sock} {
+ global URL robotsRunning
+
close $sock
after cancel $URL($sock,cancel)
- while {1} {
+
+ foreach v [array names URL $url,*] {
+ unset URL($v)
+ }
+
+ incr robotsRunning -1
+ RobotStart
+}
+
+proc RobotStart {} {
+ global URL
+ global robotsRunning robotsMax idleTime
+
+ # puts "RobotStart"
+ while {1} {
set url [RobotFileNext unvisited]
if {![string length $url]} {
- break
+ return
+ }
+ incr robotsRunning
+ if {[string compare $url wait] == 0} {
+ after $idleTime RobotRR
+ return
}
set r [RobotGetUrl $url {}]
if {!$r} {
- return
+ if {$robotsRunning >= $robotsMax} return
} else {
- RobotFileUnlink unvisited $URL($url,host) $URL($url,path)
- }
+ incr robotsRunning -1
+ if {![RobotFileExist bad $URL($url,hostport) $URL($url,path)]} {
+ set outf [RobotFileOpen bad $URL($url,hostport) $URL($url,path)]
+ RobotFileClose $outf
+ }
+ RobotFileUnlink unvisited $URL($url,hostport) $URL($url,path)
+ }
}
- incr robotMoreWork -1
}
proc headSave {url out} {
upvar $pathx path
puts "Ref url = $url href=$href"
+
+ if {[string first { } $href] >= 0} {
+ return 0
+ }
+ if {[string length $href] > 256} {
+ return 0
+ }
+ if {[string first {?} $url] >= 0 && [string first {?} $href] >= 0} {
+ return 0
+ }
# get method (if any)
if {![regexp {^([^/:]+):(.*)} $href x method hpath]} {
set hpath $href
if {![string length $surl]} {
set surl /
}
- set ok 0
- foreach domain $domains {
- if {[string match $domain $host]} {
- set ok 1
- break
+ if {[info exist domains]} {
+ set ok 0
+ foreach domain $domains {
+ if {[string match $domain $host]} {
+ set ok 1
+ break
+ }
}
- }
- if {!$ok} {
- return 0
- }
+ if {!$ok} {
+ return 0
+ }
+ }
} else {
regexp {^([^\#]*)} $hpath x surl
- set host $URL($url,host)
+ set host $URL($url,hostport)
}
if {![string length $surl]} {
return 0
set surl $dpart/$surl
}
}
- set c [split $surl /]
- set i [llength $c]
- incr i -1
- set path [lindex $c $i]
- incr i -1
- while {$i >= 0} {
- switch -- [lindex $c $i] {
+ set surllist [split $surl /]
+ catch {unset path}
+ set pathl 0
+ foreach c $surllist {
+ switch -- $c {
.. {
- incr i -2
- if {$i < 0} {
- set i 0
+ if {$pathl > 0} {
+ incr pathl -1
+ set path [lrange $path 0 $pathl]
}
}
- . {
- incr i -1
- }
- default {
- set path [lindex $c $i]/$path
- incr i -1
+ . {
+
+ }
+ default {
+ incr pathl
+ lappend path $c
}
}
}
- regsub -all {~} $path {%7E} path
- set ok 1
- if {[info exists URL($host,robots)]} {
- foreach l $URL($host,robots) {
- if {[string first [lindex $l 1] $path] == 0} {
- set ok [lindex $l 0]
- break
- }
- }
+ if {$pathl} {
+ set path [join $path /]
+ } else {
+ set path ""
}
+ regsub -all {~} $path {%7E} path
set href "$method://$host$path"
- puts "Ref href = $href, ok=$ok"
- return $ok
+ puts "Ref href = $href"
+ return 1
}
proc RobotError {url code} {
puts "Bad URL $url, $code"
set fromurl {}
- if {[RobotFileExist unvisited $URL($url,host) $URL($url,path)]} {
- set inf [RobotFileOpen unvisited $URL($url,host) $URL($url,path) r]
- set fromurl [gets $inf]
- close $inf
- }
- RobotFileUnlink unvisited $URL($url,host) $URL($url,path)
- if {![RobotFileExist bad $URL($url,host) $URL($url,path)]} {
- set outf [RobotFileOpen bad $URL($url,host) $URL($url,path)]
- puts $outf "URL=$url $code"
- puts $outf "Reference $fromurl"
+ set distance -1
+ if {[RobotFileExist unvisited $URL($url,hostport) $URL($url,path)]} {
+ set inf [RobotFileOpen unvisited $URL($url,hostport) $URL($url,path) r]
+ RobotReadRecord $inf fromurl distance
+ RobotFileClose $inf
+ }
+ RobotFileUnlink unvisited $URL($url,hostport) $URL($url,path)
+ if {![RobotFileExist bad $URL($url,hostport) $URL($url,path)]} {
+ set outf [RobotFileOpen bad $URL($url,hostport) $URL($url,path)]
+ RobotWriteRecord $outf $fromurl $distance
RobotFileClose $outf
}
}
puts "Redirecting from $url to $tourl"
+ set distance {}
set fromurl {}
- if {[RobotFileExist unvisited $URL($url,host) $URL($url,path)]} {
- set inf [RobotFileOpen unvisited $URL($url,host) $URL($url,path) r]
- set fromurl [gets $inf]
+ if {[RobotFileExist unvisited $URL($url,hostport) $URL($url,path)]} {
+ set inf [RobotFileOpen unvisited $URL($url,hostport) $URL($url,path) r]
+ RobotReadRecord $inf fromurl distance
RobotFileClose $inf
}
- if {[catch {RobotFileUnlink unvisited $URL($url,host) $URL($url,path)}]} {
- puts "unlink failed"
- exit 1
- }
- if {![RobotFileExist bad $URL($url,host) $URL($url,path)]} {
- set outf [RobotFileOpen bad $URL($url,host) $URL($url,path)]
- puts $outf "URL=$url to $tourl $code"
- puts $outf "Reference $fromurl"
+ if {![RobotFileExist bad $URL($url,hostport) $URL($url,path)]} {
+ set outf [RobotFileOpen bad $URL($url,hostport) $URL($url,path)]
+ RobotWriteRecord $outf $fromurl $distance
RobotFileClose $outf
}
if {[RobotHref $url tourl host path]} {
- if {![RobotFileExist unvisited $host $path]} {
- puts "Mark as unvisited"
- set outf [RobotFileOpen unvisited $host $path]
- puts $outf $code
- RobotFileClose $outf
+ if {![RobotFileExist visited $host $path]} {
+ if {![RobotFileExist unvisited $host $path]} {
+ set outf [RobotFileOpen unvisited $host $path]
+ RobotWriteRecord $outf $fromurl $distance
+ RobotFileClose $outf
+ }
+ } else {
+ set olddistance {}
+ set inf [RobotFileOpen visited $host $path r]
+ RobotReadRecord $inf oldurl olddistance
+ RobotFileClose $inf
+ if {[string length $olddistance] == 0} {
+ set olddistance 1000
+ }
+ if {[string length $distance] == 0} {
+ set distance 1000
+ }
+ puts "distance=$distance olddistance=$olddistance"
+ if {[expr $distance < $olddistance]} {
+ set outf [RobotFileOpen unvisited $host $path]
+ RobotWriteRecord $outf $tourl $distance
+ RobotFileClose $outf
+ }
}
}
+ if {[catch {RobotFileUnlink unvisited $URL($url,hostport) $URL($url,path)}]} {
+ puts "unlink failed"
+ exit 1
+ }
}
proc RobotTextHtml {url out} {
- global URL
+ global URL maxDistance
+ set distance 0
+ if {$maxDistance < 1000 && [info exists URL($url,dist)]} {
+ set distance [expr $URL($url,dist) + 1]
+ }
htmlSwitch $URL($url,buf) \
title {
puts $out "<title>$body</title>"
}
puts $out {></meta>}
} body {
- regsub -all -nocase {<script.*</script>} $body {} abody
+ regsub -all -nocase {<script([^<]|(<!.*>))*</script>} $body {} abody
regsub -all {<[^\>]+>} $abody {} nbody
puts $out "<documentcontent>"
puts $out $nbody
puts $out "</documentcontent>"
- } a {
+ } -nonest a {
if {![info exists parm(href)]} {
puts "no href"
continue
}
- if {1} {
- set href $parm(href)
+ if {[expr $distance <= $maxDistance]} {
+ set href [string trim $parm(href)]
if {![RobotHref $url href host path]} continue
puts $out "<cr>"
puts $out "<identifier>$href</identifier>"
puts $out "<description>$body</description>"
puts $out "</cr>"
+
+ if {![RobotFileExist visited $host $path]} {
+ set olddistance 1000
+ if {![RobotFileExist bad $host $path]} {
+ if {[RobotFileExist unvisited $host $path]} {
+ set inf [RobotFileOpen unvisited $host $path r]
+ RobotReadRecord $inf oldurl olddistance
+ RobotFileClose $inf
+ }
+ } else {
+ set olddistance 0
+ }
+ if {[string length $olddistance] == 0} {
+ set olddistance 1000
+ }
+ if {[expr $distance < $olddistance]} {
+ set outf [RobotFileOpen unvisited $host $path]
+ RobotWriteRecord $outf $url $distance
+ RobotFileClose $outf
+ }
+ } elseif {[string compare $href $url]} {
+ set inf [RobotFileOpen visited $host $path r]
+ RobotReadRecord $inf xurl olddistance
+ close $inf
+ if {[string length $olddistance] == 0} {
+ set olddistance 1000
+ }
+ if {[expr $distance < $olddistance]} {
+ puts "OK remarking url=$url href=$href"
+ puts "olddistance = $olddistance"
+ puts "newdistance = $distance"
+ set outf [RobotFileOpen unvisited $host $path]
+ RobotWriteRecord $outf $url $distance
+ RobotFileClose $outf
+ }
+ }
+ }
+ } -nonest area {
+ if {![info exists parm(href)]} {
+ puts "no href"
+ continue
+ }
+ if {[expr $distance <= $maxDistance]} {
+ set href [string trim $parm(href)]
+ if {![RobotHref $url href host path]} continue
+ puts $out "<cr>"
+ puts $out "<identifier>$href</identifier>"
+ puts $out "<description></description>"
+ puts $out "</cr>"
+
if {![RobotFileExist visited $host $path]} {
+ set olddistance 1000
if {![RobotFileExist bad $host $path]} {
- if {[catch {set outf [RobotFileOpen unvisited $host $path]} msg]} {
- puts "--- Error $msg"
- exit 1
- }
- puts $outf $url
+ if {[RobotFileExist unvisited $host $path]} {
+ set inf [RobotFileOpen unvisited $host $path r]
+ RobotReadRecord $inf oldurl olddistance
+ RobotFileClose $inf
+ }
+ } else {
+ set olddistance 0
+ }
+ if {[string length $olddistance] == 0} {
+ set olddistance 1000
+ }
+ if {[expr $distance < $olddistance]} {
+ set outf [RobotFileOpen unvisited $host $path]
+ RobotWriteRecord $outf $url $distance
+ RobotFileClose $outf
+ }
+ } elseif {[string compare $href $url]} {
+ set inf [RobotFileOpen visited $host $path r]
+ RobotReadRecord $inf xurl olddistance
+ close $inf
+ if {[string length $olddistance] == 0} {
+ set olddistance 1000
+ }
+ if {[expr $distance < $olddistance]} {
+ puts "OK remarking url=$url href=$href"
+ puts "olddistance = $olddistance"
+ puts "newdistance = $distance"
+ set outf [RobotFileOpen unvisited $host $path]
+ RobotWriteRecord $outf $url $distance
RobotFileClose $outf
}
}
proc RobotsTxt {url} {
global agent URL
- set v URL($URL($url,host),robots)
+ RobotsTxt0 URL(URL($url,hostport),robots) $URL($url,buf)
+}
+
+proc RobotsTxt0 {v buf} {
+ global URL agent
set section 0
- foreach l [split $URL($url,buf) \n] {
- puts $l
- if {[regexp {([-A-Za-z]+):[ \t]*([^\#]+)} $l match cmd arg]} {
+ foreach l [split $buf \n] {
+ if {[regexp {([-A-Za-z]+):[ ]*([^\# ]+)} $l match cmd arg]} {
puts "cmd=$cmd arg=$arg"
- switch $cmd {
- User-Agent {
+ switch -- [string tolower $cmd] {
+ user-agent {
if {$section} break
set pat [string tolower $arg]*
set section [string match $pat $agent]
}
- Disallow {
+ disallow {
if {$section} {
puts "rule [list 0 $arg]"
lappend $v [list 0 $arg]
}
}
- Allow {
+ allow {
if {$section} {
puts "rule [list 1 $arg]"
lappend $v [list 1 $arg]
global URL
puts $out "<documentcontent>"
- puts $out $URL($url,buf)
+ regsub -all {<} $URL($url,buf) {\<} content
+ puts $out $content
puts $out "</documentcontent>"
if {![string compare $URL($url,path) /robots.txt]} {
proc Robot200 {url} {
global URL domains
- puts "Parsing $url"
- set out [RobotFileOpen visited $URL($url,host) $URL($url,path)]
+ set out [RobotFileOpen raw $URL($url,hostport) $URL($url,path)]
+ puts -nonewline $out $URL($url,buf)
+ RobotFileClose $out
+
+ set out [RobotFileOpen visited $URL($url,hostport) $URL($url,path)]
puts $out "<zmbot>"
+
+ set distance 1000
+ if {[RobotFileExist unvisited $URL($url,hostport) $URL($url,path)]} {
+ set inf [RobotFileOpen unvisited $URL($url,hostport) $URL($url,path) r]
+ RobotReadRecord $inf fromurl distance
+ RobotFileClose $inf
+ }
+ set URL($url,dist) $distance
+ puts $out "<distance>"
+ puts $out " $distance"
+ puts $out "</distance>"
headSave $url $out
+ puts "Parsing $url distance=$distance"
switch $URL($url,head,content-type) {
text/html {
- RobotTextHtml $url $out
+ if {[string length $distance]} {
+ RobotTextHtml $url $out
+ }
}
text/plain {
RobotTextPlain $url $out
puts $out "</zmbot>"
RobotFileClose $out
# puts "Parsing done"
- RobotFileUnlink unvisited $URL($url,host) $URL($url,path)
+ RobotFileUnlink unvisited $URL($url,hostport) $URL($url,path)
}
proc RobotReadContent {url sock binary} {
global URL
+ puts "RobotReadContent $url"
set buffer [read $sock 16384]
set readCount [string length $buffer]
if {$readCount <= 0} {
Robot200 $url
- RobotRestart $sock
+ RobotRestart $url $sock
} elseif {!$binary && [string first \0 $buffer] >= 0} {
Robot200 $url
- RobotRestart $sock
+ RobotRestart $url $sock
} else {
# puts "Got $readCount bytes"
set URL($url,buf) $URL($url,buf)$buffer
puts "RobotReadHeader $url"
if {[catch {set buffer [read $sock 2148]}]} {
RobotError $url 404
- RobotRestart $sock
+ RobotRestart $url $sock
}
set readCount [string length $buffer]
if {$readCount <= 0} {
RobotError $url 404
- RobotRestart $sock
+ RobotRestart $url $sock
} else {
# puts "Got $readCount bytes"
set URL($url,buf) $URL($url,buf)$buffer
regexp {^HTTP/([0-9.]+)[ ]+([0-9]+)} $headbuf x version code
set lines [split $headbuf \n]
foreach line $lines {
- if {[regexp {^([^:]+):[ ]+(.*)} $line x name value]} {
+ if {[regexp {^([^:]+):[ ]+([^;]*)} $line x name value]} {
set URL($url,head,[string tolower $name]) [string trim $value]
}
}
switch $code {
301 {
RobotRedirect $url $URL($url,head,location) 301
- RobotRestart $sock
+ RobotRestart $url $sock
}
302 {
RobotRedirect $url $URL($url,head,location) 302
- RobotRestart $sock
+ RobotRestart $url $sock
}
200 {
if {![info exists URL($url,head,content-type)]} {
}
default {
RobotError $url $code
- RobotRestart $sock
+ RobotRestart $url $sock
}
}
}
}
}
-proc RobotSockCancel {sock url} {
+proc RobotSockCancel {url sock} {
puts "RobotSockCancel sock=$sock url=$url"
RobotError $url 401
- RobotRestart $sock
+ RobotRestart $url $sock
}
proc RobotConnect {url sock} {
puts $sock "User-Agent: $agent"
puts $sock ""
flush $sock
- set URL($sock,cancel) [after 60000 [list RobotSockCancel $sock $url]]
+ set URL($sock,cancel) [after 30000 [list RobotSockCancel $url $sock]]
}
proc RobotNop {} {
}
proc RobotGetUrl {url phost} {
- global URL
+ global URL robotsRunning
flush stdout
- puts "---------"
- puts $url
- if {![regexp {([^:]+)://([^/]+)([^ ]*)} $url x method hostport path]} {
+ puts "RobotGetUrl --------- robotsRunning=$robotsRunning url=$url"
+ if {![regexp {([^:]+)://([^/]+)(.*)} $url x method hostport path]} {
return -1
}
if {![regexp {([^:]+):([0-9]+)} $hostport x host port]} {
}
set URL($url,method) $method
set URL($url,host) $host
- set URL($url,port) $port
+ set URL($url,hostport) $hostport
set URL($url,path) $path
set URL($url,state) head
set URL($url,buf) {}
+
+ if {[string compare $path /robots.txt]} {
+ set ok 1
+ if {![info exists URL($hostport,robots)]} {
+ puts "READING robots.txt for host $hostport"
+ if {[RobotFileExist visited $hostport /robots.txt]} {
+ set inf [RobotFileOpen visited $hostport /robots.txt r]
+ set buf [read $inf 32768]
+ close $inf
+ } else {
+ set buf "User-agent: *\nAllow: /\n"
+ }
+ RobotsTxt0 URL($hostport,robots) $buf
+ }
+ if {[info exists URL($hostport,robots)]} {
+ foreach l $URL($hostport,robots) {
+ if {[string first [lindex $l 1] $path] == 0} {
+ set ok [lindex $l 0]
+ break
+ }
+ }
+ }
+ if {!$ok} {
+ puts "skipped due to robots.txt"
+ return -1
+ }
+ }
if [catch {set sock [socket -async $host $port]}] {
return -1
}
set agent "zmbot/0.0"
if {![catch {set os [exec uname -s -r]}]} {
set agent "$agent ($os)"
- puts "agent: $agent"
}
+puts "agent: $agent"
+
proc bgerror {m} {
global errorInfo
puts "BGERROR $m"
puts $errorInfo
}
-set robotMoreWork 0
+set robotsRunning 0
set robotSeq 0
set workdir [pwd]
+set idleTime 60000
+
+set i 0
+set l [llength $argv]
-if {[llength $argv] < 2} {
- puts "Tclrobot: usage <domain> <start>"
- puts " Example: '*.indexdata.dk' http://www.indexdata.dk/"
+# For testing only
+if {0} {
+ set url "http://www.sportsfiskeren.dk/sportsfiskeren/corner/index.htm"
+ set href "../../data/../../data2/newsovs.asp?Mode=5"
+
+ set URL($url,path) /sportsfiskeren/corner/index.htm
+ set URL($url,hostport) www.sportsfiskeren.dk
+ RobotHref $url href host path
+ exit 0
+}
+
+if {$l < 2} {
+ puts {tclrobot: usage [-j jobs] [-i idle] [-c count] [-d domain] [url ..]}
+ puts " Example: -c 3 -d '*.dk' http://www.indexdata.dk/"
exit 1
}
-set domains [lindex $argv 0]
-foreach site [lindex $argv 1] {
- incr robotMoreWork
- if [RobotGetUrl $site {}] {
- incr robotMoreWork -1
- puts "Couldn't process $site"
+while {$i < $l} {
+ set arg [lindex $argv $i]
+ switch -glob -- $arg {
+ -j* {
+ set robotsMax [string range $arg 2 end]
+ if {![string length $robotsMax]} {
+ set robotsMax [lindex $argv [incr i]]
+ }
+ }
+ -c* {
+ set maxDistance [string range $arg 2 end]
+ if {![string length $maxDistance]} {
+ set maxDistance [lindex $argv [incr i]]
+ }
+ }
+ -d* {
+ set dom [string range $arg 2 end]
+ if {![string length $dom]} {
+ set dom [lindex $argv [incr i]]
+ }
+ lappend domains $dom
+ }
+ -i* {
+ set idleTime [string range $arg 2 end]
+ if {![string length $idleTime]} {
+ set idleTime [lindex $argv [incr i]]
+ }
+ }
+ default {
+ set href $arg
+ if {[RobotHref http://www.indexdata.dk/ href host path]} {
+ if {![RobotFileExist visited $host $path]} {
+ set outf [RobotFileOpen unvisited $host $path]
+ RobotWriteRecord $outf href 0
+ RobotFileClose $outf
+ }
+ }
+ }
}
+ incr i
+}
+
+if {![info exist domains]} {
+ set domains {*}
}
+if {![info exist maxDistance]} {
+ set maxDistance 50
+}
+if {![info exist robotsMax]} {
+ set robotsMax 5
+}
+
+puts "domains=$domains"
+puts "max distance=$maxDistance"
+puts "max jobs=$robotsMax"
+
+RobotStart
-while {$robotMoreWork} {
- vwait robotMoreWork
+while {$robotsRunning} {
+ vwait robotsRunning
}