X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=dcdot.tcl;h=a29e6f5414c136b99ce6073d8aed5b641179d9db;hb=563c2ed58045d7cd4c977af162f13d1fa308b4a5;hp=aeb4f91484a9e635a060ee5d3ff9f6e66bc8c51a;hpb=d8234df96ab8fb03ed71f6358f7211ebe725b495;p=tclrobot.git

diff --git a/dcdot.tcl b/dcdot.tcl
index aeb4f91..a29e6f5 100755
--- a/dcdot.tcl
+++ b/dcdot.tcl
@@ -1,5 +1,5 @@
 #!/usr/bin/tclsh 
-# $Id: dcdot.tcl,v 1.1 2000/12/07 20:16:11 adam Exp $
+# $Id: dcdot.tcl,v 1.3 2000/12/08 22:46:53 adam Exp $
 #
 
 proc RobotRestart {} {
@@ -11,28 +11,14 @@ proc RobotRestart {} {
 proc RobotTextHtml {url} {
     global URL
     
-    set head 0
-    htmlSwitch $URL($url,buf) \
-        title {
-	    set URL($url,title) $body
-        } -nonest meta {
-	    set scheme {}
-	    if {[info exist parm(scheme)]} {
-		set scheme $parm(scheme)
-		unset parm(scheme)
-	    }
-	    if {[info exist parm(name)]} {
-		if {[info exist parm(content)]} {
-		    set URL($url,meta,$parm(name),$scheme) $parm(content)
-		    unset parm(content)
-		}
-		unset parm(name)
-	    }
-        } a {
-            if {[info exists parm(href)]} {
-		lappend URL($url,links) $parm(href)
-	    }
-	}
+    set b $URL($url,buf)
+    set e {<[mM][eE][tT][aA][^>]*>}
+    catch {unset $URL($url,meta)}
+    while {[regexp -indices $e $b i]} {
+        set meta [string range $b [lindex $i 0] [lindex $i 1]]
+	lappend URL($url,meta) $meta
+        set b [string range $b [lindex $i 1] end]
+    }
 }
 
 proc Robot200 {url} {
@@ -113,7 +99,6 @@ proc RobotReadHeader {url sock} {
 		    }
 		}
 		default {
-		    Robot404 $url
 		    close $sock
 		    RobotRestart
 		}
@@ -157,14 +142,7 @@ proc RobotGetUrl {url phost} {
     return 0
 }
 
-if {![llength [info commands htmlSwitch]]} {
-    set e [info sharedlibextension]
-    if {[catch {load ./tclrobot$e}]} {
-	load tclrobot$e
-    }
-}
-
-set agent "zmbot/0.0"
+set agent "dcdot.tcl/0.0"
 if {![catch {set os [exec uname -s -r]}]} {
     set agent "$agent ($os)"
 }
@@ -185,8 +163,13 @@ proc RobotGetDCDOT {url} {
 if {$argc == 1} {
     set url [lindex $argv 0]
     RobotGetDCDOT $url
-    set mask {,meta,[Dd][Cc]\.*}
-    foreach a [array names URL $url$mask] {
-	puts "URL($a) = $URL($a)"
+    set mask {,meta}
+    if {[info exist URL($url,meta)]} {
+        foreach m $URL($url,meta) {
+	    puts $m
+        }
     }
-}
\ No newline at end of file
+    foreach v [array names URL $url,head,*] {
+	puts "$v = $URL($v)"
+    }
+}